CONNECTIONS BETWEEN THE DIFFERENTIATION OF HIGHER EDUCATION PARTICIPATION AND THE DISTRIBUTION OF OCCUPATIONAL STATUS

ABSTRACT

This article addresses how the type of higher education participation an individual has chosen is related to his/her success on the labour market after graduation. In the first part of the analysis, the ideal types of ‘from school-via higher education-to work’ transitions are formed with the help of cluster analysis. In the second part of the study, the relationship between the type of transition and the socio-economic prestige of the occupation obtained after graduation is analysed through regression analysis. Data used in the analysis comprise educational and labour market histories of graduates with master's degree in seven European countries. Results indicate that there are significant variations among European countries in terms of the most common types of transitions. In addition, the study shows that the socio-economic level reached soon after graduation does in fact depend on the type of the transition, independently of other individual characteristics.

Keywords:

• higher education participation
• pan-European
• labour market success
• students in higher education
• transition from school to work

Notes

¹For a descriptive analysis of the CHEERS data, see Kivinen and Nurmi (2003).

²The survey respondents were asked how many months they spent (a) in education, (b) in employment, (c) in childrearing or family care, (d) in military or civilian service, or (e) unemployed during the time between becoming eligible to apply for HE and the actual beginning of the studies. Correspondingly, the survey respondents were asked how many months they spent in different activities during their studies. The different alternatives were the same as those listed above, except for employment, which was divided into relevant and not relevant work in relation to the field and the level of studies. The respondents were free to report any number of activities of the alternatives listed above. The different activities may have been simultaneous rather than successive but that can not be seen from the data.

³Cases in the data are not grouped into categories based on their actual cluster, but the results of the cluster analysis have been used in an indirect way, as the basis for making the classification rules. This method was used in order to avoid an overlapping of some of the clusters. Since the classification rules are not totally exclusive, some of the cases in the data could be grouped into more than just one of the categories. This applies to individuals who have reported two or more activities lasting long enough to fall within the limits of two or more categories. These kinds of cases have been grouped into a single category according to the activity which lasted the longest. The number of these ambiguous cases is nevertheless small.

⁴The censored models considered in this study are the so-called Tobit models, which are regression models for left censored data assuming a normally distributed error term (see e.g., Pindyck and Rubinfeld, 1991). Parameter estimates of the censored regression models, as used in this article, can be interpreted in the same manner as estimates of the conventional OLS models.